Insights recipe: First reply time by event

Have more questions? Submit a request

53 Comments

  • Amy Dee
    Comment actions Permalink

    Hi Oliver! It looks like you posted this question over in the community as well. There are some notes on the post, and you should have an email soon to follow up in a ticket.

    Watch for that email, and happy reporting!

    0
  • Paul
    Comment actions Permalink

    Hi Amy,

     

    I am having a small problem with this. I have check and re-checked, just to make sure I wasn't doing something wrong. The majority of my results are negative, as in -9.95 for example.

     

    Thanks

    0
  • Amy Dee
    Comment actions Permalink

    Hi Paul! There are two things I would check up front.

    First, just to confirm, are you using Ticket updated (minutes) or Ticket updated at (minutes)? The two look similar, but they do different things. Ticket updated (minutes) returns the timestamp for any given update, which is what you need for this recipe. Ticket updated at (minutes) returns the most recent update timestamp, which isn't helpful here.

    Second, do you get a lot of tickets through inbound ticket sharing agreements? The creation and update timestamps for shared tickets can get a little skewed, since updates are coming from two different accounts. You may need to report on those tickets separately.

    If you're still having trouble finding the cause for these discrepancies, I recommend sending in a support ticket. Please include your draft report and metrics, along with a few example tickets that return negative numbers. We'll take a closer look.

    I hope this helps! Happy reporting!

    0
  • Jaime Bonacif
    Comment actions Permalink

    Hello Amy,

    The comparison metric finds the first response time for each ticket, but is it possible to change the metric so that I know how long each agent worked on the ticket since the last change during business hours?

    For example: Agent 1 responded to a ticket one hour after creation and forwarded the ticket to Agent 2 at the end of the work day. Agent 2 answered the ticket the other day, one hour after arriving at work. Thus, the two agents have ONE TIME of response time.

    With this information, I believe it is possible to see the average response time per group within business hours.

    Can you help me please?

    0
  • Jessie Schutz
    Comment actions Permalink

    Hi Jaime! Welcome to the Community!

    Have you checked out the Time Tracking app? I believe that  will help you get the data you're looking for.

    0
  • Jeremy Wilson
    Comment actions Permalink

    Hello Amy,

    I've attempted to recreate your report but I must be doing something wrong.  I need to find an average response time per agent (preferably in minutes if possible) on a daily/weekly/monthly basis.  The initial response time should be the the time it takes an agent to enter a public reply after assigning themselves a ticket.  

    When I re-create your report, my first reply time by event (hrs) [Avg] column is showing very large negative numbers (-687.27) per agent for any given day. 

    Any help would be greatly appreciated.

    0
  • Amy Dee
    Comment actions Permalink

    Hi Jeremy! There are a lot of moving parts in this recipe, which means there are a lot of picky little things that can go wrong.

    The quickest way to figure out what happened is to go through some specific examples. I'm going to create a ticket for you, so we can look at your report and review some individual tickets.

    Watch for my email, and happy reporting!

    0
  • Kelsey Hales
    Comment actions Permalink

    Love this article, this was very helpful and taught me a bit about how some of the MAQL custom metrics work. Something I'm trying to track is the total # of first replies an agent sent vs non-first replies (our handle time expectation for a first response is much more brief than a second or third response). How do I isolate this? I've added Ticket Id to the How, and that pulls what appears to be correct data, but I can't seem to get just a total number. Any ideas where to start on this?

    0
  • Mariko Yano
    Comment actions Permalink

    Is it possible to get first reply time in business hours of a ticket after they are assigned to a certain group?

    Ex. A ticket has been re-assigned from tier 1 to tier 2 and a tier 2 agent replies. Tier 1 previously entered a public comment. 

    Is it possible to capture Tier 2's reply time from the time the ticket was assigned til tier 2 replies in business hours? 

    0
  • Dan Cooper
    Comment actions Permalink

    Hi Mariko, 

    Unfortunately it's not possible to show first reply time for replies that occur later in a ticket. You could potentially adapt your workflow to capture this by having your tier 1 create a new ticket for your tier 2 agents, but you'd give up the total time that the customer has been waiting for that entire interaction. 

    0
  • Mariko Yano
    Comment actions Permalink

    Thanks Dan. I tried making a public comment an internal comment, then add another public comment. That action still didn't update my first reply time. Is that expected?

    0
  • Dan Cooper
    Comment actions Permalink

    I'm not sure if the private to public comment would work, but it's an interesting idea.  The insights refresh time isn't instant (it's 1 hour on our Enterprise plan) so you may need to wait to see if that sticks.  As far as I'm aware, I haven't heard anyone else try what you call out, so if it works in your testing, share it.  It would be great to give people an option if they need one. 

    0
  • Amy Dee
    Comment actions Permalink

    Hi Mariko! Hi Dan! There are a lot of good tips here!

    If you're interested in reporting on a specific duration that isn't in a default metric, like the time from a group change to a public agent comment, you can use the timestamps for those events. We have a recipe article here: Insights recipe: Duration between two or more ticket events in minutes. This type of recipe gives you a lot more flexibility, but it also has a lot of moving parts. Make sure you take your time and audit your results.

    Please note: Insights can't see your Zendesk schedule, so all custom durations will be in calendar hours. Only the default Zendesk metrics and business hour SLA targets can be reported on in business hours.

    As for switching a comment from public to private, I don't think it can change the default Zendesk metrics. Those are recorded at the time the reply is made, so they should still hold. However, it may change a custom duration, since attributes like Public Comment show the current value. You have me curious, though, so I may test that out as well.

    I hope this helps! Happy reporting!

    0
  • Mariko Yano
    Comment actions Permalink

    Hi Amy!

    I've already followed that recipe :) I tried doing it that way but struggling with getting our first reply time in business hours using excel.

    I've listed down our business hours below (we have APAC and Seattle Support hours). The gaps in support hours make it a little challenging for me. Had we been doing 24/7 support, this would be easy. :)

    sunday 5pm-11:59pm

    Monday: 12am-2am, 8am-11:59pm

    Tuesday: 12am-2am; 8am-11:59pm

    Wednesday: 12am-2am; 8am-11;59pm

    Thursday: 12am-2am; 8am-11:59pm

    Friday: 12am-2am; 8am-5pm

     

     

    0
  • Dan Cooper
    Comment actions Permalink

    I'm really glad someone came in to debunk my own searches (and you have a super awesome article on it already)!  Thanks Amy! 

    0
  • Amy Dee
    Comment actions Permalink

    @Dan - You're welcome! That's definitely a fun article for building crazy reports. As a data nerd, I highly recommend losing an afternoon with it.

    @Mariko - Insights can't see your schedule, and Zendesk only records business hours on default metrics and business hour SLAs. There is no way to do this conversion automatically with native Zendesk tools. Once you get into custom durations like this, you would need to do a lot of complicated and repetitive math outside Insights to filter for business hours.

    That said, the values in the Ticket updated (minutes) fact are just epoch timestamps in minutes. There may be apps or programs out there that can do this sort of conversion for any epoch start and end times. If you find something like that, you could export the timestamps on their own and calculation the duration separately.

    Please note - most epoch timestamps are recorded in seconds, and Insights uses minutes. You'd probably need to multiply your results by 60 to be compatible with an epoch-based app or script.

    I hope this helps! Happy reporting!

    0
  • Brandal Henao
    Comment actions Permalink

    Hi Amy 

    I tried using this recipe but it's not taking the correct values in some cases and I don't know why :S 

    This is what I have: 

    where the second filter is the one recommended by you adding the empty values (tried the default recipe and it didn't work as expected either .. 


    The organization with org_tags is based on your other recipe https://support.zendesk.com/hc/en-us/articles/228202567-Reporting-on-user-and-organization-tags

    Any ideas why it's not working as expected?, the average should be around 0-2 for all weeks .. 

    Thank you !



    0
  • Amy Dee
    Comment actions Permalink

    Hi Brandal! There are a lot of moving parts in this report covering a wide range of datasets. At first glance, I suspect there are a few high outliers in those first two weeks, or there are edge case tickets sneaking through the filters.

    You can dig in to see where the numbers are coming from by drilling into individual tickets. GoodData has more information in their documentation on drilling into reports (use the report-level steps).

    In your case, set the first reply time metric to drill into the Ticket Id. That will allow you to click on those numbers for W27 and W28 and see which tickets were used in the calculation. That should give you a better idea of what's happening.

    This type of troubleshooting usually gets into specific ticket examples. If you have trouble narrowing it down with the list of tickets, I recommend sending in a support ticket. 

    I hope this helps! Happy reporting!

    0
  • Ceren Has
    Comment actions Permalink

    Is it possible to build a "% change in first reply time since previous week" key metric? If yes, how? :)

    0
  • Olivier Pascal Netatmo
    Comment actions Permalink

    Hi Amy

    I am far away to be an expert but I would like to know if I have to follow the same steps to get the reply time for new tickets and for open tickets?

    Thank you

    0
  • Amy Dee
    Comment actions Permalink

    To all the awesome folks following this comment section -

    There's a new version of this recipe! You can find the details here: First reply time by event (version 2). The new version builds the filtering into the metric itself, so you don't need the three-part numeric range filter. This makes the metric much easier to use. Give the new version a try in your next report!

    To Ceren - 

    The percent change report you described won't work with this version of the recipe, since you can't see the previous week with the numeric range filter in place. Fortunately, the new version of the recipe does allow you get those results, since the filters are all built in.

    Try building the new version (available here), then use it in the template from this recipe article: Time-over-time reporting. You'd just need to use the new first reply metric instead of # Tickets, and Week (Mon-Sun)/Year (Event) instead of Month/Year (Ticket Created).

    To Olivier -

    This recipe (old and new version) doesn't include any filter for the ticket status beyond excluding deleted tickets. It should give you the same results regardless of whether the ticket is in new or open status.

     

    I hope this helps! Happy reporting, everybody!

    0
  • annie jasmine
    Comment actions Permalink

    Amy,

     

    How can I get the below report in Overview into our insights? I do not find this report in the list of reports in good data. I tried using "Tickets by First Reply' but that extracts the count of tickets rather than %.

    Thanks 

    0
  • Amy Dee
    Comment actions Permalink

    Hi Annie! We have a different recipe article showing how to recreate that exact graph. You can find it here: Reviewing SLA performance in Insights.

    That recipe article shows you how to use the native first reply time metric to find each range, then how to use those ranges to find percentages. It's also much simpler than the event-based recipe in the article above, so it should be easier to set up.

    I hope this helps! Happy reporting!

    0

Please sign in to leave a comment.

Powered by Zendesk